EP2036272A1

EP2036272A1 - Method for maintaining synchronisation in a radio communications system

Info

Publication number: EP2036272A1
Application number: EP07786757A
Authority: EP
Inventors: Mieszko Chmiel; Jaroslaw Niewczas
Original assignee: Nokia Siemens Networks GmbH and Co KG
Current assignee: Nokia Solutions and Networks GmbH and Co KG
Priority date: 2006-06-21
Filing date: 2007-06-18
Publication date: 2009-03-18
Also published as: WO2007147808A1; US8311000B2; US20100067497A1

Abstract

Method for maintaining synchronisation in a radio communications system According to the invention, a method for maintaining synchronisation of a user terminal (UE) in a radio communication system is proposed, wherein resources reserved for transmissions of random access signals are devided into at least one first portion for contention-based communication, and at least one second portion of said resources for contention- free communication, and wherein a user terminal (UE) is assigned at least one resource of the portion assigned to the contention-free communication, for transmitting signals for maintaining synchronisaton.

Description

Method for maintaining synchronisation in a radio communications system

The invention relates to a method for maintaining synchronisation in a radio communications system. Specifically, the invention relates to synchronous random access operation in mixed contention-based and contention-free modes for uplink time synchronisation and/or power control.

In radio communications systems, situation when a mobile sta^¬ tion has temporary no data to transmit and/or receive, i.e. it is inactive, arise. However, it is desirable that the mo- bile station can quickly restart its transmission and/or reception. Such short reactivation time of less than 50 ms is defined as one of the fundamental requirements of currently standardized E-UTRA, see for example the technical specifica^¬ tion 3GPP TS 25.913. It is assumed that user terminals (UE - User Equipment) will be kept synchronised with the network for a certain time even when not being in an active state anymore. More specifically, the UE, e.g. a dormant user ter^¬ minal in a so called RRC Connected state, shall be kept up^¬ link (UL) time synchronised in order to allow for a fast re- start of transmissions and avoid time consuming non-synchro^¬ nised random access procedures. During such inactivity time, the UE may be configured by the network with DRX/DTX (Discontinuous Reception/Transmission) cycles. Moreover, due to efficiency reasons it may be desirable that after some time of inactivity, which may depend on the used service, priority etc., the UE is allowed to lose the UL time synchronisation, and in this case in order to obtain a timing advance or to request resources, the UE must use so called non-synchronised random access mechanisms. During a random access, it is possible for the user terminal to communicate with a network' s base station (Node B) in both non-synchronised, i.e. with no time synchronisation and lack of precise power control, and synchronised mode. Synchronised mode is both time- and power-controlled via closed-power loop. The present invention described below deals with syn^¬ chronised mode of operation and describes how to efficiently keep inactive user terminals uplink time synchronised.

According to the invention, the problem of the inefficiency of non-synchronised contention-based random access is solved. It is proposed to maintain uplink synchronisation of dormant user terminals using synchronised, contention-free random ac- cess, while also allowing for "on-demand" user terminal re^¬ source requests using synchronised contention-based random access .

In radio communications systems presently known, different mechanisms for initial and random access exist, which are discussed in the following.

According to the GSM standard, uplink transmissions from dif^¬ ferent user terminals are time aligned at the base station: • Initial Access: contention-based access burst containing synchronisation sequence, and small amount of data are sent by the user terminal so that the base station can estimate and if necessary adjust the user terminal's up^¬ link timing advance. An access burst uses a guard period to take into account initial time uncertainty due to an unknown propagation delay. The transmission timing of access bursts is based on the timing of received downlink signal. • Circuit switched voice: the timing advance is controlled by the base station. During voice spurts the uplink timing measurements are based on normal burst voice packets or the dedicated Slow Associated Control Channel (SACCH) . During voice gaps, i.e. when DTX is used at the user terminal, the uplink timing measurements are based either on periodically send SID (silence indicator de^¬ scription) frames or on the SACCH. The time advance up^¬ date interval is constant i.e. every four multi-frames (480 ms) .

• Packet switched data (GPRS) : the timing advance is con^¬ trolled by the base station. When the user terminal is allocated uplink resources and is transmitting, the tim^¬ ing advance measurements are based on the transmitted data packet (s) . The timing advance update interval is every four multi-frames (480 ms ) . When the user terminal is not transmitting or has no allocated uplink re^¬ sources, the timing advance is based on contention-free (possibly non-synchronised, i.e. the guard time is still used) access bursts sent by the user terminal once every eight multi-frames (960 ms ) .

According to the WCDMA (Wideband CDMA or UTRA FDD) standard:

• Initial access: the user terminal uses the contention- based random access channel, i.e. preamble part with power open loop power control and power ramping, for initial access to establish connection or request re^¬ sources, but not to obtain uplink time advance.

• Uplink transmissions from different user terminals are not time aligned at the base station. The timing of up^¬ link transmissions at the user terminal has a fixed re^¬ lation to the timing of the received downlink (DL) signal. Therefore, UL signals of user terminals with dif- ferent propagation delays are not time aligned at the base station.

According to the so called WiMAX standard IEEE 802.16: • A specific number of sub-channels are allocated for ran^¬ dom access, both initial and periodic (used for timing and power adjustments). Access type is contention-based only .

According to the current proposals for 3GPP Long Term Evolu^¬ tion (LTE), received uplink signals (sub-frames) from differ^¬ ent user terminals are time aligned at the base station:

• Initial access: contention-based non-synchronised random access bursts are used to obtain uplink time advance and to request resources. The non-synchronised random access bursts contain at least a signature sequence and possi^¬ bly a message part. Currently, one or two-step proce^¬ dures are considered, see for example the technical specification TS 25.814. These bursts are used during initial access bit they may be also used when the user terminal lost uplink time synchronisation or while in non-synchronised handover. The TDM/FDM (Time/Frequency Division Multiplex) structure of the non-synchronised random access channel (RACH) is considered, i.e. bursts use time/frequency resources which are orthogonal to re^¬ sources used for other scheduled data transmissions. Currently, the assumptions is that due to a required guard time one sub-frame (0.5 ms) of three or multiple of three resource units (75 sub-carriers) will be re- served for non-synchronised random access in every radio frame (10 ms ) .

• Synchronised random access: it is furthermore assumed that contention-based synchronised random access bursts are used to request uplink resources when the user ter^¬ minal is uplink time synchronised. Such bursts occupy significantly less time/frequency resources than non- synchronized random access bursts, because no guard time is needed. As an example, this may be one DFT-OFDM sym^¬ bol (66.66 us) with one or multiple frequency resource units (25 sub-carriers) every two sub-frames.

According to one aspect of the invention, it is proposed to reserve a first part of available synchronous random access resources for contention-based communication in order to per^¬ form user terminal resource requests, while a second part of the available resources are used for contention-free communi^¬ cation. In the following, this latter part of the resources is called dummy synchronized random access bursts.

It is assumed that traffic on the contention-free part is controlled by the base station, wherein each user terminal is assigned a unique transmit opportunity, e.g. a slot, at regu- lar pre-determined intervals, e.g. one every second or half a second. With such fixed transmission pattern, no downlink resources would be needed for scheduling. The contention-free dummy random access bursts are used to maintain sufficient uplink timing synchronisation and power control, so that the user terminals are able to reactivate almost instantly with high probability.

According to a further aspect of the invention, the interval for transmitting dummy random access bursts is variably de- termined based on e.g. the current user terminal's speed or capability to move.

According to a further aspect of the invention, dummy random access bursts are used not only to measure and possibly ad- just the user terminal' s timing advance and/or transmit power via the base station, but also for carrying some signaling information like e.g. downlink CQIs (channel quality indica^¬ tor) .

According to a further aspect of the invention, dummy random access bursts used for transmissions of timing control and/or power control and/or CQI are retransmitted in case of failed reception at the base station.

According to a further aspect of the invention, contention- based random access bursts are used for instantly requesting resources, and possibly for retransmission of contention-free random access bursts with resource requests. Advantageously, appropriate splitting of the synchronised random access re^¬ sources allows for a substantial reduction in the overall re^¬ quired overhead by exploiting trunking gain. The percentage of such contended traffic would be small, possibly in the range of 5-10% of the synchronised random access resources.

According to a further aspect of the invention, dummy random access bursts are transmitted together with uplink reference symbols, i.e. symbols known to the receiver, to allow uplink channel sounding in order to maintain up-to-date information about the user terminal's channel condition at the scheduler. Such reference symbols may also be used for channel estima^¬ tion to improve demodulation and/or detection of the synchronised random access bursts.

In contrast to known algorithms based on non-synchronised contention-based random access, which would make it difficult to enable a reconnection within 50 ms, the method according to the invention is flexible and allows for more predictable and guaranteed ability to reconnect with only a limited num^¬ ber of resources assigned to it.

As an example, in case of contended-mode, whether synchro- nised or not, a significant number of resource would have to be allocated to keep the probability of collisions of signals at the receiver low. If for example, the average RACH traffic is 5 messages per 10 ms (in a cell) , one would need to pro^¬ vide at least 20 unique opportunities within that 10 ms pe- riod, to keep the probability of collisions below 25%, or >50 opportunities for a probability of collisions below 10%. Such resource allocation would significantly diminish the overall capacity of the system. In contrast, with a scheduled conten^¬ tion-free access according to the invention, only 5 opportu- nities would be needed.

Moreover, the separation of synchronised RACH resources into contended and contention-free resources reflects more accu^¬ rately the nature of traffic in the random access channel, wherein a significant portion is regular and thus predict^¬ able, like e.g. RACH sequences transmitted by user terminals to maintain synchronisation with the base station. Instead, the use of a contented mode would not be efficient for such traffic because of collision risks, delays arising from such collisions and required additional overhead. The use of the known contention-free mode provides transmit opportunities at regular intervals, e.g. in the range of 0.25 s to 5 s, which would be far too long. Since only a small portion of the overall traffic would require the contended mode, the addi- tional overhead would not be significant. According to the invention, it is therefore proposed to separate the synchro^¬ nised resources, thus allowing both ways of access. Alternative or additional mechanisms for tracking uplink tim^¬ ing and/or transmit power of user terminals are discussed in the following.

According to one aspect, CQI reports (transmitted via outband signalling in uplink to the base station) of downlink channel conditions determined by the user terminal could be used to track the user terminal's timing and/or transmit power. There is a need for transmitting such CQI reports whenever the user terminal receives data in downlink from the base station. In^¬ stead, if no data is sent in uplink or received in downlink, CQI reports may not be transmitted or transmitted with a longer interval compared with the case when the downlink is active, see for example the technical report TR 25.903 relat- ing to the so called continuous connectivity.

Furthermore, uplink reference signals, e.g. pilot symbols, may be used to track uplink timing and/or transmit power. However, such resources assigned to pilots are limited and more related to detection, demodulation and channel sounding procedures of active users.

According to another aspect of the present invention, which might be regarded as being independent from the above de- scription, a position of a random access burst within a frame is disclosed.

According to the current proposals for 3GPP Long Term Evolu^¬ tion (LTE), is assumed that synchronised random access bursts are transmitted in one or several so called long blocks (LB) within a sub-frame. The sub-frame structure shown in fig. 1 is e.g. disclosed in the technical specification TR 25.814, V7.0.0, figure 9.1.1-4. Such long blocks are protected by guard intervals called cyclic prefix (CP) . There are 8 cyclic prefixes per sub-frame but only one of them is mandated to be longer than the others, in order to allow filter ramp- ups/ramp-downs .

According to this further aspect of the invention, it is pro^¬ posed to use such long block protected by a longer cyclic prefix for transmitting synchronised random access bursts. According to fig. 1, this could be e.g. the long block LB#1 preceded by a longer cyclic prefix. Of course, at present there is no definition of which of the long blocks will be preceded by the longer cyclic prefix, therefore, the inven^¬ tion is not limited to aforesaid example.

Depending on the system bandwidth, the length of a long block is 66.67 us and the regular cyclic prefix is approx. 4.13

...3.65 us, while the longer cyclic prefix is approx. 4.39...7.81 us long.

By transmitting synchronised random access bursts in blocks which are preceded by a longer guard interval, additional protection is achieved because especially the time synchroni- saton of the above described dummy random access bursts is not as accurate, i.e. less than the duration of the cyclic prefix, as it is in active mode when data is being transmit- ted in the uplink. In active mode, the accuracy is within a fraction of a cyclic prefix.

Assuming an equal probability of the user terminal UE moving towards or away from the base station Node B, it is more likely that the move away form the base station Node B will trigger handover signalling sooner that the move towards the base station Node B, as it is illustrated in fig. 2. For this reason, it would be more beneficial to put the signals used for maintaining uplink synchronisation at the longer cyclic prefix, as shown in fig. 3. Loosing synchronisation due to shorter propagation delays is more likely than loosing synchronisation due to longer propagation delays, because the latter one, on average, will sooner trigger handover signal- ing, which in turn will is used to keep or adjust timing advance. However, placing signals used for maintaining uplink synchronisation before the longer cyclic prefix would also be advantageous compared to placing such signals before normal cyclic prefixes.

Although it would in general be sufficient to use only one long block per sub-frame transmitted every number of sub- frames for tracking uplink timing, if more consecutive time resources within a sub-frame are used for these signals, the solution to protect that signal with the longer cyclic prefix is still applicable.

Claims

1. Method for maintaining synchronisation of a user terminal (UE) in a radio communication system, wherein resources re- served for transmissions of random access signals are devided into at least one first portion for contention-based communi^¬ cation, and at least one second portion of said resources for contention-free communication, and wherein a user terminal (UE) is assigned at least one resource of the portion as- signed to the contention-free communication, for transmitting signals for maintaining synchronisaton .

2. Method according to claim 1, wherein the user terminal (UE) is in an inactive state when using the assigned at least one resource.

3. Method according to claim 1 or 2, wherein the user terminal (UE) uses the at least one resource for a predetermined time interval.

4. User terminal (UE), comprising means for realising the method according to claim 1.

5. Base station (Node B) or a radio communications system, comprising means for realising the method according to claim

1.

6. Radio communications system, comprising of at least one base station (Node B) and at least one user terminal (UE) , each comprising means for realising the method according to claim 1.